Injecting CO2 with huff-n-puff process is becoming an efficient technique to overcome some productivity damage and enhanced gas recovery in depleted gas reservoirs. Several numerical simulations are required to estimate the gas recovery factor obtained from CO2 huff-n-puff injection. However, numerical simulations are very expensive and time-consuming processes. This study aims to develop a predictive model to estimate the additional gas recovery factor by CO2 huff-n-puff injection in depleted carbonate gas reservoirs. To create the predictive model, a base case reservoir model was constructed based on a real carbonate gas field data. Subsequently, ranges of reservoir and CO2 huff-n-puff operation variables that affect the gas recovery factor were determined from published literature data. To obtain optimal spacing and cover all of the variable space, space filling design together with Latin hypercube method was then employed to construct 100 simulation cases prepared using CMG-GEM. Using multiple regression method, a linear predictive model was developed to estimate the additional gas recovery factor obtained from CO2 huff-n-puff injection in depleted carbonate gas reservoirs. The novelty of this model is in the ability to quick estimation of the additional gas recovery factor by CO2 huff-n-puff process using typical reservoir and huff-n-puff operation parameters. © 2018 Society of Petroleum Engineers. All rights reserved.